RAG PDF Query Application

Overview

This project provides a Retrieval-Augmented Generation (RAG) solution for processing PDF documents and answering user queries through a web interface. It combines AI-powered responses, PDF management, and session handling. Users can upload PDFs, ask questions, and interact with the application via a modern web interface.

Data Source: The application works with user-uploaded PDF documents and leverages OpenAI's API for intelligent query responses.

Table of Contents

• Project Features
• Project Architecture
• Technologies Used
• Running the Application
• Project Structure
• Usage Examples
• Secure Deployment on AWS Lightsail
• Contact

Project Features

• PDF Upload and Management: Users can upload and switch between multiple PDF documents.
• AI-Powered Querying: Ask questions about the content of uploaded PDFs and receive intelligent answers.
• Session and Cache Handling: Uses Redis to manage user sessions and cache query results.
• Interactive Web Interface: Built with Flask and JavaScript for a responsive user experience.

Project Architecture

Below is the high-level architecture, including document upload, processing, querying, and visualization.

1. Document Upload: Users upload PDF files via the web interface.
2. Backend Processing: The backend extracts text, manages documents, and handles AI queries.
3. Query Handling: Integrates OpenAI API for RAG-based answers.
4. Session Management: Redis stores session data and caches results.
5. Frontend Visualization: Flask and JavaScript display results and manage user interaction.

---

Technologies Used

• Python: Core programming language.
• Flask: Web application framework for the frontend.
• Custom Backend: Handles PDF processing and AI queries.
• Redis: For session and cache management.
• OpenAI API: For AI-powered query responses.
• JavaScript: Enhances frontend interactivity.
• pip: Dependency management.

Running the Application

1. Clone the Repository:

   git clone https://github.com/Carlos93U/ai_query_pdf.git
   cd ai_query_pdf

Project Structure

ai_query_pdf/
├── assets
│     ├── demo.gif
│     └── img.png
├── backend                         # PDF processing and AI query logic
│     ├── app.py                    # Backend entry point
│     ├── Dockerfile                # Dockerfile for backend
│     ├── requirements.txt
│     ├── uploads
│     │     └── key_word.pdf
│     └── vectorstore
├── compose.yml                     # Docker compose for project
├── .env                            # Environment configuration
│
├── frontend                        # Flask web interface
│     ├── app.py                    # Frontend entry point
│     ├── Dockerfile                # Dockerfile for frontend
│     ├── requirements.txt
│     ├── static
│     │     ├── script.js
│     │     └── style.css
│     └── templates
│         └── index.html
├── LICENSE
└── README.md                       # Project documentation

Usage Examples

Users can upload PDF documents, ask questions, and view AI-generated answers directly in the web interface.

Secure Deployment of app-chat on AWS Lightsail via SSH

🏗 Step 0 — Create the Instance in Lightsail

1. Go to AWS Lightsail.
2. Click Create instance.
3. Configure:
   • Region: the closest to you.
   • Platform: Linux/Unix.
   • Blueprint: Ubuntu 22.04 LTS.
   • Plan: according to your resource needs.
   • Name: app-chat.
4. Click Create instance and wait for it to start.
5. Note the Public IP of the instance.

---

🧩 Step 1 — Connect via SSH

1. Download the Lightsail private key for your region, e.g., LightsailDefaultKey-us-east-1.pem and move it to ~/.ssh/ directory.
2. Set read-only permissions:

chmod 400 ~/.ssh/LightsailDefaultKey-us-east-1.pem

3. Connect to the instance (replace <PUBLIC_IP> with your instance's IP):

ssh -i ~/.ssh/LightsailDefaultKey-us-east-1.pem ubuntu@<PUBLIC_IP>

---

🧩 Step 2 — Prepare the VM and Install Docker

Inside the VM via SSH:

# Update and prepare
sudo apt update -y
sudo apt upgrade -y

# Install Docker and Docker Compose
sudo apt install -y docker.io docker-compose

# Enable and start Docker
sudo systemctl enable --now docker

# Add 'ubuntu' user to docker group
sudo usermod -aG docker ubuntu

# Verify installation
docker --version
docker compose version || docker-compose --version
docker run --rm hello-world

Important note:

Remember to log out and back in for the docker group changes to take effect.

---

🧩 Step 3 — Create the Project on the VM

mkdir ~/app-chat
cd ~/app-chat
mkdir frontend backend

---

🧩 Step 4 — Generate OpenAI Key and Configure .env

1. Go to OpenAI and generate your API key.
2. On your local machine, create a .env file with:

OPENAI_API_KEY="YOUR_REAL_KEY"
BACKEND_URL=http://backend:8000
REDIS_URL=redis://redis:6379
VECTORSTORE_PATH=/app/vectorstore
UPLOAD_FOLDER=/app/uploads
SECRET_KEY=ultra_secret

3. Upload .env to the VM:

scp -i ~/.ssh/LightsailDefaultKey-us-east-1.pem .env ubuntu@<PUBLIC_IP>:/home/ubuntu/app-chat/

Alternatively, edit it directly on the VM:

nano ~/app-chat/.env

---

🧩 Step 5 — Upload App Files

From your local machine:

scp -i ~/.ssh/LightsailDefaultKey-us-east-1.pem -r ./frontend ubuntu@<PUBLIC_IP>:/home/ubuntu/app-chat/
scp -i ~/.ssh/LightsailDefaultKey-us-east-1.pem -r ./backend ubuntu@<PUBLIC_IP>:/home/ubuntu/app-chat/
scp -i ~/.ssh/LightsailDefaultKey-us-east-1.pem docker-compose.yml ubuntu@<PUBLIC_IP>:/home/ubuntu/app-chat/

If you don't have the files yet, you can create them directly on the VM using nano or cat.

---

🌍 Step 6 — Open Ports in Lightsail

1. Go to your app-chat instance in Lightsail.
2. Tab Networking → Firewall → Add another.
3. Add TCP rules:

Port	Usage
5000	Frontend
8000	Backend

Save the changes.

---

🚀 Step 7 — Run the Project

From the VM:

cd ~/app-chat
docker compose up -d

• This starts the containers in detached mode.
• Verify they are running:

docker ps

Expected output:

CONTAINER ID   IMAGE                PORTS
a1b2c3d4e5f6   app-chat_frontend    0.0.0.0:5000->5000/tcp
b2c3d4e5f6a7   app-chat_backend     0.0.0.0:8000->8000/tcp
...            ...                  ...

---

🧭 Step 8 — Test in the Browser

1. Open in your browser:

http://<PUBLIC_IP>:5000

2. Verify the frontend works and the backend responds correctly.

---

🗑 Step 9 — Clean Up Resources to Avoid Unnecessary Costs

1. Stop running containers:

docker compose down

2. Optional: remove images if not needed:

docker system prune -a

3. Delete the Lightsail instance:
   • Go to AWS Lightsail.
   • Select your app-chat instance.
   • Click Delete and confirm.

This ensures you don't incur ongoing costs for unused resources.

Contact

For questions or feedback, please contact:

Email: huillcas.juan3@gmail.com
GitHub: Carlos93U

---